Most galaxies have a supermassive black hole lurking at their centers, but a galaxy 10.5 billion light-years away looks like it might have two—and the pair may be set to crash together in just 21 years. If the observations are confirmed, the duo would be the closest known set of binary black holes, and their imminent collision would offer scientists an unprecedented chance to watch extreme physics in action. [Video: When Black Holes Collide]

The potential black holes are impossible to glimpse directly. Besides being black—which is to say, invisible—they are far too distant from Earth and too close to one another to make out with telescopes. But scientists found what they think is the telltale signature of a pair of these behemoths circling together. The apparent black holes reside in what’s known as a quasar—a galaxy that is unleashing a torrent of light as mass gets gobbled by the gigantic black hole at its center. The light from most quasars flickers randomly as slightly more or less mass falls in. But if two black holes are at the core, rather than one, their orbital motions will perturb the gas around them in an orderly way, causing the light to rise and fall periodically.

Tingting Liu, a graduate student at the University of Maryland, College Park, analyzed the light pouring from 316 quasars as seen by the Pan-STARRS 1 (Panoramic Survey Telescope and Rapid Response System) Medium-Deep Survey, a scan of the sky undertaken by the Pan-STARRS1 telescope on Mount Haleakala in Hawaii. She looked for periodic patterns in the light and a quasar with the unwieldy name PSO J334.2028+01.4075 turned out to have the best evidence for a recurring signal. The black hole or holesthere have an estimated combined mass between three billion and 30 billion times the mass of the sun. “This candidate that we present is at such a compact separation that they are actually in the process of merging,” says Suvi Gezari, leader of Liu’s research group and a co-author on a paper announcing their find that has been accepted for publication in the The Astrophysical Journal Letters. In fact, the impending impact will seem to happen even sooner for the black holes themselves. In their reference frame, they are due to converge in just seven years. Because of a phenomenon called cosmological time dilation, due to the expansion of the universe, the crash will seem to take place in 21 years from our point of view on Earth. “This was a fortuitous catch,” says Stuart Shapiro, a theoretical astrophysicist at the University of Illinois at Urbana–Champaign who was not involved in the research. “We don’t know of another candidate that’s anywhere near that close [to merging].” But, he adds, “I take it with a grain of salt, and even if it doesn’t pan out there are probably many more on the horizon.”

Astronomers expect to find more binary black holes with the rise of Pan-STARRS1 and other surveys like it that can monitor such systems over time and look for periodic light variations. Shapiro uses computer simulations to predict what happens when giant black holes fuse. To be able to observe one in space would offer him and other theorists a precious reality check of their calculations. “When two black holes get sufficiently close, we think they plunge together suddenly and they merge,” Shapiro says. “That plunge and merger will give rise to a burst of gravitational waves, and that burst will then diminish for awhile as the merged remnant rings down, like a bell.” The bell-like vibrations should send out spirals of gravitational waves, which are ripples that stretch and shrink the fabric of spacetime.

Such ripples could be detectable to so-called pulsar timing arrays, which use spinning stars called pulsars as natural clocks to look for disturbances from gravitational waves. Pulsars spin very regularly and sweep jets of light out like lighthouse beams. If a gravitational wave passed through space, their light might come off schedule. By comparing the timing of many pulsars across the cosmos, astronomers could identify when and where a gravitational wave came from. “Right now we don’t actually have the sensitivity to detect this system [PSO J334.2028+01.4075] but we’re finding more and more of these through various surveys so this is potentially the tip of the iceberg,” says Xavier Siemens of the University of Wisconsin–Milwaukee, who leads the North American Nanohertz Observatory for Gravitational Waves (NANOGrav) pulsar timing array project. “This system is potentially something we could detect with next-generation radio telescopes,” he adds, “with things like the Square Kilometer Array” due to start up around 2025 in South Africa and Australia.

Whether or not PSO J334.2028+01.4075 is truly a binary black hole is still an open question. George Djorgovski, an astronomer at Caltech who recently found another candidate double–back hole quasar, is not convinced. “I am skeptical about their analysis and claims,” he says. The Pan-STARRS1 survey only observed the system in question on a handful of occasions, so the apparent variation in light could turn out to be chance, he cautions. Furthermore, the likelihood of finding such a rare event in such a relatively small initial sample size (316 quasars) is low. “The chances of catching one pair at such a close separation as they claim—that would merge within several years—is probably well less than one in a million,” he adds. “Liu et al would have to be very, very lucky.”

Astronomers should not have to wait too long to find out. Liu and her collaborators calculated that the quasar’s light fluctuates over a regular period of about 542 days, meaning that forthcoming data should soon either confirm or reject the pattern. “It’s a really easy thing to test the persistence of this periodic fluctuation in the future,” Gezari says. “And if there are two black holes that are starting to in-spiral toward one another, then their period should be getting smaller and smaller. We can actually look for that change as a function of time.” If the pattern holds up, astronomers just may get a front-row seat to one of nature’s most extreme events.